Adaptive Local Phase Approximations and Global Unwrapping

The paper introduces a new modulo-2pi phase denoising algorithm based on local polynomial approximations. The zero and first order approximations of the phase are calculated in sliding windows of varying size. The former is used for point- wise adaptive window size selection, while the latter is used for filtering the phase in the obtained windows. For unwrapping, we input the PUMA unwrapping algorithm [1] with the denoised wrapped phase obtained with the proposed approach. Simulation shows that this technique enables strong noise attenuation while preserving image details.